Wayne Davison [Thu, 29 Apr 2004 19:34:31 +0000 (19:34 +0000)]
Restore old behavior of logging most daemon errors instead of sending
them to the user. This should eventually be improved to duplicate
some of these messages to the user to keep them informed about what
went wrong.
Wayne Davison [Tue, 27 Apr 2004 03:53:28 +0000 (03:53 +0000)]
- Tweaked the bullet char so that vim can auto-format the items more
easily (and then used vim to re-flow the items).
- Added a mention about the daemon security fix.
- Added the 2.6.1 release date.
Wayne Davison [Thu, 22 Apr 2004 09:58:15 +0000 (09:58 +0000)]
Initialize the new debug_type string in each exclude list and output
it in a couple extra debug messages. When we free an exclude list,
make sure we don't clear the debug_type string.
Wayne Davison [Sat, 17 Apr 2004 17:14:12 +0000 (17:14 +0000)]
Changed the dev handling for -H back to using an opaque 64-bit
integer instead of trying to transfer it as separate major & minor
values. Since the value is not interpreted by the receiving side
(just compared for equality), this is a safer way to go.
Wayne Davison [Wed, 14 Apr 2004 23:33:40 +0000 (23:33 +0000)]
Our exclude lists are now comprised of a base structure with a head
and a tail pointer pointing to a linked list of items (rather than
an array of pointers that has to be reallocated all the time).
Wayne Davison [Wed, 14 Apr 2004 23:33:37 +0000 (23:33 +0000)]
Added a "next" pointer to "exclude_struct" and added a new structure
for exclude lists, exclude_list_struct, that holds the head/tail
pointers for the list.
Wayne Davison [Wed, 14 Apr 2004 21:59:41 +0000 (21:59 +0000)]
The check_exclude() and report_exclude_result() functions now take a
description string that describes what kind of an include/exclude is
happening (which only gets output when verbose is set to 2 or more).
Wayne Davison [Mon, 12 Apr 2004 17:42:32 +0000 (17:42 +0000)]
- add_exclude() now takes a flag int and can optionally word-split
items and/or ignore +/- prefixes.
- add_exclude_file() changed to take the same flags as add_exclude().
- add_exclude_line() was removed since add_exclude() can now do its job.
- make_exclude() now takes a size-limited string without a +/- prefix.
- add_cvs_excludes() was changed to word-split the items that come from
a .cvsignore file, and to ignore all +/- prefixes in all the ignore
sources. Since CVS doesn't check if its ignore items are directories,
the dir-names (e.g. SCCS, CVS, etc.) no longer have a trailing slash
(making our excludes match CVS's ignores more exactly).
Wayne Davison [Mon, 12 Apr 2004 16:52:40 +0000 (16:52 +0000)]
- Use dev_t instead of DEV64_T.
- Changed rdev_high to rdev_major.
- Use major(), minor(), and makedev() to manipulate device numbers.
- Send the major and minor values separately in protocol 28 and beyond.
Wayne Davison [Mon, 12 Apr 2004 16:52:37 +0000 (16:52 +0000)]
- Got rid of DEV64_T and changed the internal device variables to dev_t.
- Changed XMIT_SAME_HIGH_RDEV to XMIT_SAME_RDEV_MAJOR.
- Added XMIT_RDEV_MINOR_IS_SMALL.
Wayne Davison [Sat, 10 Apr 2004 10:31:11 +0000 (10:31 +0000)]
Fixed a bug where an exclude name that got sent over the wire could get
an extra "- " or "+ " parsed off the start of the name (i.e. we have to
quote excluded names that start with those strings with an extra "- "
at the start).
Wayne Davison [Wed, 7 Apr 2004 23:09:14 +0000 (23:09 +0000)]
If --daemon was specified, exit the option-parsing code before
adjusting various default values (the real defaults will get
set when the client's options arrive).
Wayne Davison [Wed, 31 Mar 2004 18:53:57 +0000 (18:53 +0000)]
Improved the include/exclude example section some more. Document
the new username default of "guest" if neither USER or LOGNAME is
set in the environment.
Wayne Davison [Mon, 29 Mar 2004 17:02:31 +0000 (17:02 +0000)]
Try to avoid running on a system that allows us to create devices
without being root (like cygwin) because rsync won't copy the
devices if it is not root.